What To Do If Your Book Was Stolen And Used To Train AI

Authors, your rights and your consent matter. Several authors are suing Meta to defend those rights. But what happens when others take advantage for the sake of getting ahead in the tech race? Here’s what you should do if your book was stolen and used to train AI.

What’s Happening:

The rise of AI use within the publishing industry has raised ethical concerns among the author community centered around quality, originality, copyright, and artistic integrity. The most recent cause for outrage features Meta Platforms at the forefront, with accusations from multiple authors stating that the company used pirated books from the pirate website Library Genesis (LibGen) to train their own AI model. Authors including Sarah Silverman, Richard Kadrey, and Ta-Nehisi Coates have filed lawsuits claiming that Meta utilized their books without permission to develop its AI models (Wired).

Recently unsealed court documents revealed that Meta’s CEO, Mark Zuckerberg approved the use of pirated materials, including the works of various authors. This raised serious concerns about the legality of using pirated content and intensified debates about ethics when it comes to using copyrighted books to train AI (The Guardian). Now several authors are suing Meta as a result.

Class-action lawsuits have been filed by several authors against Meta, OpenAI, Microsoft, and other AI companies, alleging copyright infringement. By filing, the authors hope to address the unauthorized use of their novels in training AI and gain rightful compensation (The Authors Guild). 

The publishing world anxiously awaits the outcomes of these cases, as they will set a significant precedent for future AI laws regarding intellectual property and technological advancement.

What to Do If Your Book Was Stolen and Used to Train AI

1. Confirm Whether Your Book Was Used:

• Investigate lists: Some AI models have been trained on datasets like Books3, which contain pirated books. These lists have occasionally been leaked or made public. You can try to find out whether your work appears in one of them.
• Search tools: Tools like Have I Been Trained let you search for your work in image datasets, but similar tools for text datasets are more limited.

2. Join Class Action Lawsuits:

• Several lawsuits have already been filed against Meta and other AI companies (e.g., OpenAI) by authors like Sarah Silverman, Paul Tremblay, and Mona Awad.
• Stay informed through groups like:
  • The Authors Guild (they’ve filed lawsuits and are pushing for AI regulation)
  • The Society of Authors in the UK
• If you’re eligible, you may be able to join or support a class action or file a claim once settlements are reached.

3. Document the Use

• Collect evidence: If you can find specific outputs or examples showing your book has been used, take screenshots, record dates, and keep links.

• This might include:

  • A chatbot summarizing your book too accurately

  • Passages or writing that mimic your unique style

  • Pirated versions of your book being scraped and used in datasets

4. Submit a Takedown Notice (DMCA):

• If you find your work in a dataset or output without permission, you can file a Digital Millennium Copyright Act (DMCA) takedown request with:
  • Meta or other AI companies
  • Hosting platforms of the dataset

5. Contact a Copyright Attorney:

• If your book was stolen and used without your consent, copyright law is on your side.
• A specialized intellectual property or publishing lawyer can:
  • Help you understand if you have a valid claim
  • Assist with sending cease-and-desist letters or filing a formal lawsuit

6. Advocate for Legal Reform:

• Support or advocate for AI regulation and authors’ rights legislation.
• Engage in campaigns organized by:
  • The Authors Guild
  • Fairly Trained (a nonprofit advocating for ethical AI training)

7. License Your Work Ethically (Optional):

If you’re open to your work being used for AI training—but on your terms—consider:

• Licensing your books through platforms like Fairly Trained or Spawning, which focus on consent-based training.
• Using metadata tags like Do Not Train (though not all companies respect this yet).

Looking for more to read? Check out Book Marketing Mistakes That Sabotage Your Sales, Author Platform: Good First Impression & Build Presence, and Find Your Ideal Reader: Market Your Novel & Create Superfans.

Happy Writing!